The Guardian ran this story, based on DfE statistics on 26 July 2013:
Under a scare headline asking “Are children becoming more violent” it took apart the Telegraph’s selective reporting of the data which had focused on a very small subset of exclusions. I think the answer to the headline was “Probably not, on balance.” So the piece’s careful emphasis on paying good attention to the data was well done.
However, the piece included a paragraph which bothered me greatly because it didn’t at all accord with my experience of working with schools and low income families on the problem of school exclusions.
I decided to take a look at the tables to which the article referred. They are here:
The two tables which refer to wealth (as far as I can see) are Tables 17 and 18. I’m pretty sure they show that wealth (or rather the lack of it) does significantly impact on exclusions. Here’s why:
The only data table I think the wealth claim can be drawn from is Table 18. But Table 18 doesn’t refer to the wealth of pupils. It is based on the deprivation level in the “super output area” in which the school is located. It tells you something, overall, about whether the school’s intake is more or less likely to include children from low or high income families, but it doesn’t tell you anything about the wealth of the excluded pupils themselves. It doesn’t tell the whole story about the composition of the school intake, either, because super output areas are quite small, but school catchment areas (especially at secondary) can be very large.
The % difference between the top and bottom deciles in Table 18 may not be huge (and I haven’t done the maths to say whether it is statistically significant), but I’m surprised that the clear evidence that schools in the wealthier upper five deciles exclude fewer children isn’t at least “journalistically significant”. Certainly I’m not sure that “wealth makes very little impact” can be truly derived from this table. At best, the claim that can be made is that there isn’t a very big difference in the % of exclusions from schools located in wealthy super output areas compared with poorer ones. (A closer look at actual catchment areas served might undermine even this claim though).
There is a table which looks at the “wealth” of individuals excluded. This is Table 17. The table doesn’t consider wealth by decile, only the binary distinction between eligible for free school meals (FSM) or not eligible. FSM eligibility is a pretty accurate proxy for poverty (in that those all who are eligible are very definitely poor) but not a complete proxy, in that it is possible to be quite poor and not eligible for FSM, or very poor and unaware of eligibility for FSM. Children from families where an adult member circulates in and out of low paid work may not have featured as an FSM child in this table. (In other words, the data on exclusions of children who are not eligible for free school meals will also include some children who are from low income families).
I think Table 17 shows clearly that children on Free School Meals are very much more likely to be excluded from school, both permanently and on a fixed term basis, and from every type of school, but most especially at secondary school. Though if other, more statistical minds want to double-check the conclusions I’ve come to, that would be great.
The available data can’t tell us everything I’d be interested in knowing about wealth/poverty and exclusions, but it does prompt two further questions which at the moment can’t be answered from the data in the tables:
A) Is it likely that of the excluded children who are not eligible for free school meals, a disproportionate number are also from low income families just above the eligibility level?
B) Is it likely that if you are eligible for free school meals you are even more likely to be excluded from a school based in a wealthier super-output area compared with a school based in a more deprived super-output area?
My experience tells me that the answer to both of those questions is likely to be yes and that this is a matter of significant social injustice. I wonder whether the raw data exists behind the tables with which it would be possible to test these hypotheses further?
To sum up, I don’t think the data on which the article seems to have been based supports the statement that wealth has very little impact on exclusions. To me, the reverse appears to be true. Wealth – or rather the lack of it – means a child is much more likely to be excluded from school.